Magnetographic Diagnostics of
Object of Investigation
Considering medical studies, many of the Human organs,
such as the heart, are sources of a weak pulsating magnetic fields. If
measured accurately, these fields can be used to diagnose heart diseases
similar to conventional an electrocardiograph, that uses electric
potentials on the Human body surface.
A database was created (by GE) to construct magnetographic
diagnostic rules. The study compiled measurements on 63 patients: 15
with the myocardial infarction, 16 with the ventricular tachycardia and
32 normal people. For every patient, 56 magnetometric sensor points were
measured and every signal was digitized with a frequency of 500 Hz.
The research goal to formulate diagnostic rules was to
distinguish between myocardial infarction, ventricular tachycardia and
normal cases. The comparative analysis between a magnetograph and
electrocardiograph was also a primary goals.
The database contained 68,000 variables that were
considerably noised and possibly a non-linear decision rule, made the
problem analytically complex. To achieve information on these questions
would have more value than using an analytical technique.
The discrimination rules for diagnostics in two types of
diseases were formulated in magnetographic and electrographic data sets.
KET demonstrated that the potential of magnetographic measurements
revealed the diagnostic success was due to the preservation of important
high frequency components in magnetographic signals. Additionally, the
newly discovered electrocariographic rules produced more accurate
results than conventional routines.
Increase diagnostic reliability in the new sources was
found in both conventional EKG measurements and magnetographic signals.
The difference between myocardial infarction and ventricular tachycardia
cases was the diagnostic errors were reduced from 16% to 2%.
Recommendations to improve the data collection process was suggested.
it was Done
A combination of the Information Analysis and
Fourier Analysis Modules were used to reduced the data set. The
results in this stage suggested a special non-linear data converter
design that would compress the original database from 68,000 to 112
variables. The Generator of Discrimination Rules was used in the
construction the diagnostic routines. Finally, the Optimization
Modules of KET were applied to optimize the rules to achieve the
finest diagnostic's quality.
The continuation of this research would allow to created
a commercial diagnostic system combining advantages of both types of
data: EKG and MKG.